Convergent and Divergent Validity of the Learning Transfer System Inventory

Page created by Andre Cole
 
CONTINUE READING
hrdq18306.qxd 8/4/07 2:46 PM Page 385

         Convergent and Divergent
         Validity of the Learning Transfer
         System Inventory

         Elwood F. Holton III, Reid A. Bates, Annette I. Bookter,
         V. Bogdan Yamkovenko

              The Learning Transfer System Inventory (LTSI) was developed to identify a
              select set of factors with the potential to substantially enhance or inhibit
              transfer of learning to the work environment. It has undergone a variety of
              validation studies, including construct, criterion, and crosscultural studies.
              However, the convergent and divergent validity of the instrument has not
              been studied. Such a study is necessary to define the nomological network
              on the constructs. This study examines the convergent and divergent validity
              of the LTSI with twenty-eight comparison measures. Results indicate mostly
              divergent relationships, further demonstrating the uniqueness of the
              LTSI constructs. By establishing the divergent relationship with other
              known constructs, the LTSI’s usefulness for transfer research is enhanced.

         In 1996, Holton and Bates developed the learning transfer system inventory
         (LTSI) as a generalized transfer climate instrument that could be used across a
         wide variety of organizations, training programs, and employees. The LTSI,
         based on evaluation theory, was developed by examining relationships and
         constructs from previous empirical research in a grounded theory-building
         approach (Holton, 1996). Holton, Bates, and Ruona (2000) indicated that
         “without minimally validated measures, the chance for substantive mis-
         specification of models, misinterpretation of findings and measurement error
         is significantly increased” (p. 6).
              Most human resource development (HRD) professionals realize that their
         organizations need learning transfer improvement, but few have an accurate
         sense of what the problem is. As a result, even those aware of strategies to
         improve transfer (Broad & Newstrom, 1992) are left with only intuition and

         HUMAN RESOURCE DEVELOPMENT QUARTERLY, vol. 18, no. 3, Fall 2007 © Wiley Periodicals, Inc.
         Published online in Wiley InterScience (www.interscience.wiley.com) • DOI: 10.1002/hrdq.1210   385
hrdq18306.qxd 8/4/07 2:46 PM Page 386

         386                                                        Holton, Bates, Bookter, Yamkovenko

         guesswork to guide them to those most likely to yield high returns. Yet most
         would agree that improving learning transfer systems requires an ability to
         accurately diagnose factors inhibiting transfer.
              The primary reason for this paradox is that until recently, no diagnostic
         tool had emerged. In recent transfer research, a wide variety of instruments
         and measures have been used, most with either questionable or unknown
         psychometric properties. As a result, neither practitioners nor researchers have
         had a well-validated, effective diagnostic instrument. This presents a key
         barrier because it is hard to change a transfer system without accurate
         diagnosis of system problems.
              The lack of a comprehensive set of factors to measure learning transfer
         climate, which are generalizable to a wide variety of organizations and employees,
         is an issue that research should continue to address. The development of a
         generalized set of transfer factors would facilitate cross-study comparison and
         eliminate redundant instrument development efforts (Holton et al., 2000).
              Ford and Weissbein (1997) conducted a review of empirical literature on
         transfer of training that updated the earlier Baldwin and Ford (1988) study.
         Their study indicated a continuing problem with instrumentation in transfer
         research. For example, nine of the studies identified used a survey design
         (Facteau, Dobbins, Russell, Ladd, & Kudisch, 1995; Ford et al., 1992;
         Quinones, Ford, Sego, & Smith, 1995; Roullier & Goldstein, 1993; Tesluk,
         Farr, Mathieu, & Vance, 1995; Tracey, Tannenbaum, & Kavanagh, 1995; Warr &
         Bunce, 1995; Xiao, 1996) that could be considered comparable to the study
         examined in this chapter. Although all of these studies calculated internal reli-
         ability coefficients, only two (Tracey et al., 1995; Warr & Bunce, 1995) used
         factor analysis as a part of the instrument validation process. The remaining
         studies used more rudimentary validation procedures, such as simple content
         analysis, to make items as specific as possible (Facteau et al., 1995) or com-
         paring intrascale reliability to interscale correlations to establish discriminant
         validity (Xiao, 1996). This lack of attention to instrument (construct) valida-
         tion is even more alarming when it is recognized that for nearly every study
         reviewed, new customized transfer climate scales were developed.
              The measurement of transfer climate constructs has been a problematic
         issue in learning transfer research. Transfer research over the past two decades
         indicates that a wide variety of instruments and measures have been used that
         often have questionable psychometric qualities or provide little evidence
         that they measure what they purport to. We believe that the development of a
         theoretically based, psychometrically sound, and generalizable set of transfer
         climate factors is imperative if learning transfer research and practice is to move
         forward. Such a set of factors could assist researchers and practitioners by
         reducing measurement error, increasing predictive accuracy, and adding to the
         understanding of the learning transfer process and the factors that affect it. It
         would also facilitate more meaningful cross-study comparisons and minimize

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 387

         Validity of the Learning Transfer System Inventory                                       387

         the need to develop “new” measures to assess analogous constructs (Holton,
         1996; Roullier & Goldstein, 1993; Tracey et al., 1995).
             For practitioners, having a valid and reliable measure such as the LTSI
         enhances transfer because they can use such an instrument to:

         • Assess potential transfer factor problems prior to conducting major learn-
           ing interventions.
         • Follow up on evaluations of existing training programs.
         • Investigate known transfer problems.
         • Target interventions designed to enhance transfer.
         • Incorporate evaluation of transfer as part of regular employee assessments.
         • Conduct needs assessment for training programs to provide transfer skills
           to supervisors and trainers.

              Thus, the development of a research-quality diagnostic instrument to
         assess critical transfer factors is important to both researchers and practition-
         ers. There is no reason that a single tool cannot be useful in both arenas.
         Indeed, there are far too many diagnostic instruments sold for HRD practice
         that have not been tested and have no known validity.
              For the past several years, our efforts have been directed at moving toward
         this goal through the development of the LTSI. The LTSI was developed to
         identify a select set of factors with the potential to substantially enhance or
         inhibit transfer of learning to the work environment. In early development of
         the LTSI, Holton, Bates, Seyler, and Carvalho (1997) factor-analyzed nine
         constructs for transfer climate. Holton et al. (2000) expanded the instrument
         by fitting the factors to an evaluation model (Holton, 1996) and including
         additional motivation-related (for example, expectancy and motivation to
         transfer), ability-related (for example, personal capacity for transfer),
         and trainee-characteristics-related factors (for example, learner readiness and
         performance self-efficacy). Exploratory factor analysis of the resulting sixteen fac-
         tors with a large heterogeneous sample provided evidence of construct validity.
              Since then, the LTSI has undergone continuing validation. Several studies
         (Bates & Holton, 2004; Holton, Chen, & Naquin, 2003) have used the LTSI
         in different organizational settings. Four studies have provided evidence of the
         cross-cultural validity of the instrument: Yamnill (2001) in Thailand; Chen,
         Holton, and Bates (2005a) in Taiwan; Khasawneh, Bates, and Holton (2004)
         in Jordan; and Bates, Kauffeld, and Holton (2005) in Germany. In addition,
         three studies provided evidence of criterion validity and suggested that several
         work environment factors measured by the LTSI, especially for interpersonal
         supports, were powerful predictors of individual job performance following
         training (Bates, Holton, & Seyler, 1997; Bates, Holton, Seyler, & Carvalho, 2000)
         and motivation to transfer (Seyler, Holton, Bates, Burnett, & Carvalho, 1998).
         Another criterion validity study of the LTSI (Ruona, Leimbach, Holton, & Bates,

                                                HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 388

         388                                                        Holton, Bates, Bookter, Yamkovenko

         2002) suggested that reaction utility might be indirectly related to performance
         and directly related to motivation to transfer.

         Ongoing Construct Validation of the LTSI
         The goal of construct validation research is to provide evidence that research
         instruments accurately measure what they purport to. Construct-related evi-
         dence focuses primarily on the instrument score as a measure of the charac-
         teristic of interest. It demonstrates the degree to which a score on a measure
         reflects the true score of the hypothetical construct (American Educational
         Research Association, 1985).
              There are three major objectives of construct validation. First, researchers
         should specify the domain of observables related to the construct. This means
         working from theory and empirical research to identify other constructs related
         to the construct of interest, such as learning transfer climate. Second, empiri-
         cal efforts are made to determine the extent to which the constructs of interest
         tend to measure the same thing, several different things, or many different
         things. Third, researchers should perform subsequent studies or experi-
         ments to determine the extent to which supposed measures of constructs are
         consistent with best guesses about the construct (Nunnally & Bernstein, 1994).
         This could, for example, include efforts to understand what other constructs
         might or might not be correlated with the constructs of interest.
              Although satisfying these three objectives can provide the complete construct
         validation data, it is rare that all are pursued in the development of measurement
         instruments. Researchers “often develop a particular measure of a construct, skip
         aspects one and two and move directly to three and try to find interesting relations
         between their measure and measures of other constructs” (Nunnally & Bernstein,
         1994, p. 87). The authors of the LTSI addressed the first two aspects by identify-
         ing and examining potential factors that are important to the transfer learning
         climate. The next logical step in the construct validation process would be to
         gather evidence of the convergent and divergent validity of the LTSI.
              Convergent validity is concerned with “demonstrating that two independent
         methods for inferring an attribute lead to similar ends” (Nunnally & Bernstein,
         1994, p. 92). In practice, convergence is often demonstrated by examining the
         extent to which measures of the same or similar variables are correlated. The
         underlying assumption is that a measure accurately represents a variable if it
         correlates highly with other measures of the same or similar variable.
              Divergent validity is concerned with the extent to which a measure is
         novel in the sense of measuring something different from that provided by
         other measures. Divergence is thus concerned with empirically establishing a
         measure’s relative uniqueness. Divergent validity is evidenced when different
         attributes of theoretical interest are not correlated to an extremely high degree;
         that is, they share little common variance (Whitley, 1996).

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 389

         Validity of the Learning Transfer System Inventory                                      389

              Convergent and divergent validity studies also present an opportunity to
         begin to map the nomological network of the LTSI. Cronbach and Meehl
         (1955) argued that one way to provide additional evidence of a construct’s
         validity is to develop a nomological network. The term nomological, derived
         from Greek, means lawful. A nomological network can therefore be thought
         of as a lawful network of relationships or linkages between constructs
         (Trochim, 1996). Thus, the development of a nomological network can pro-
         vide evidence of construct validity to the extent that predictions such as cor-
         relations from a formal theoretical network containing the construct or
         measure of interest are confirmed.
              Two objectives guided this study. First, because no nomological network
         for learning transfer climate has been identified, one objective was to begin to
         build such a network by relating LTSI constructs to a set of theoretically sound
         comparison constructs. Second, we wanted to provide evidence of the con-
         vergent and divergent validity of the constructs measured by the LTSI by exam-
         ining the extent to which expected correlations between LTSI measures and
         comparison measures are supported. Our research was therefore framed by the
         following questions:

          1. What theoretically based, psychometrically valid comparison measures
             of constructs in the nomological network of the LTSI can be used to
             examine the convergent and divergent validity of the LTSI?
          2. What are the convergent and divergent associations between the LTSI
             scales and the comparison measures identified in question one?

         Transfer-of-Training Research
         The concept of training transfer is not new in the literature. Baldwin and Ford
         (1988) conducted a thorough review of the literature on transfer of training
         based on publications cited in major works on organizational training. They
         defined training transfer as the degree to which trainees apply to their jobs the
         knowledge, skills, behaviors, and attitudes they gained in training (Baldwin
         and Ford, 1988). Quinones, Sego, Ford, and Smith (1995) described transfer
         in simple terms of individuals exhibiting the behaviors learned in training back
         at the workplace. However, transfer of training is a complex process. Baldwin and
         Ford (1988) define it in terms of training-input factors, training outcomes,
         and conditions of transfer. Generalization of material learned on the job and
         maintenance of material over a period of time are considered conditions of
         transfer. That is, a trainee is expected to generalize the concepts learned during
         training and apply them to real-life work situations beyond the training context
         and maintain this behavior over a long period of time. In other words, transfer
         of training is observed in terms of generalization and maintenance behav-
         iors. The quality and the extent to which maintenance and generalization

                                               HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 390

         390                                                        Holton, Bates, Bookter, Yamkovenko

         behaviors are observed are largely determined by training outcomes or the
         amount of original learning that occurs on the job and retention of the mate-
         rial provided during training. In turn, the training outcomes are dependent on
         the combination of training-input factors. The input factors are subdivided into
         three sections: training design, trainee characteristics, and work-environment
         characteristics.
              Training-transfer reviews and articles (Baldwin & Ford, 1988; Bates, 2003;
         Ford & Weissbein, 1997; Bates et al., 2005) define transfer as a progression of
         events from pretraining experiences to the acquisition of cognitive knowledge
         and skills, to the capability to apply new learning to job-related tasks, to the
         application of learning to tasks and activities beyond those initially targeted by
         the training. Thus, by properly measuring and creating the most favorable
         combination of input factors, one may influence training outcomes and, even-
         tually, training transfer.
              The transfer of training is affected by a number of factors that can be clas-
         sified into three categories: trainee characteristics, training design, and work
         environment. Baldwin and Ford (1988) describe the trainee characteristics fac-
         tors in terms of ability, personality, and motivation. Holton et al. (2000) also
         include prior experiences and efficacy beliefs as constituents of trainee char-
         acteristics category. Personality factors have been described in detail by Bar-
         rick and Mount (1991) as affecting job performance. As one of the criteria of
         job performance, training proficiency is specifically affected by openness to
         experience, conscientiousness, and extraversion. A wide range of cognitive,
         psychomotor, and physical ability constructs may also transfer task perfor-
         mance. Holton et al. (2000) indicate that a set of fifty descriptor constructs for
         ability characteristics that influence task performance has been developed by
         Fleishman and Mumford (1989). Other personality characteristics like locus
         of control (Kren, 1992), job involvement (Noe & Schmitt, 1986), and organi-
         zational commitment (Mathieu & Zajac, 1990) are said to affect training-
         related motivation. Self-efficacy as a social learning concept has been suggested
         to affect training transfer through confidence in the ability to perform trained
         tasks (Ford & Weissbein, 1997). Ford, Quinones, Sego, and Sorra (1992)
         determined that opportunities to perform trained tasks are differential and in
         part determined by trainees’ self-efficacy and cognitive ability, along with
         supervisor attitudes and work group support.
              Training design has been researched probably more than any other factors
         affecting transfer (Baldwin and Ford, 1988). Baldwin and Ford (1988) describe
         training design research as centered on four basic principles: identical ele-
         ments, general principles, stimulus variability, and conditions of practice. The
         principle of identical elements calls for identical stimulus and response
         elements in transfer settings. General principles indicate that a trainee should
         be taught not just applicable skills but also general rules and theoretical
         principles that underlie training concept. Stimulus variability is the notion that
         positive transfer is maximized when a variety of relevant training stimuli

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 391

         Validity of the Learning Transfer System Inventory                                      391

         (that is, different examples rather than just one example) are employed.
         Finally, conditions of practice include feedback, massed or distributed train-
         ing, and overlearning. Baldwin and Ford (1988) provide the detailed descrip-
         tions of each of these principles. Essentially research suggests that training
         tasks should be similar to transfer tasks or tasks that employees are asked to
         do on the job using new skills or knowledge (Goldstein and Musicante, 1986).
         Also, Bates et al. (1997) argue that training content has to be consistent with
         job requirements. It is essential that training tasks be similar to transfer tasks
         (Goldstein & Musicante, 1986).
              In 1988, Baldwin and Ford stated that research on the work environment
         was limited at that time. Ten years later, Ford and Weissbein (1997) found that
         more effort had been devoted to a greater understanding and measurement of
         work environment in which the trainee was supposed to transfer the new
         knowledge and skills. There are essentially two key situations or environments
         that a person works within: the training environment and the transfer envi-
         ronment. The researchers suggest that trainee characteristics and transfer
         environment interact in the application of the knowledge and skills brought
         to the job. Therefore, if the transfer environment is favorable and the trainee
         possesses necessary characteristics, he or she is likely to be more motivated to
         transfer training to the job.
              Noe and Schmitt(1986) linked environmental favorability to pretraining
         motivation and transfer of training skills. The environmental favorability com-
         prises task constraints and perceived social support for training. Roullier
         and Goldstein (1993) suggested that transfer climate comprises situational cues
         (reducing the employee’s workload so new skills could be applied at work) and
         consequences (rewards and reinforcement). Baldwin and Ford (1988)
         described several sources of social support: top management, supervisor sup-
         port, peer support, and support by subordinates. Facteau et al. (1995) pro-
         vided empirical evidence for the positive relationship between pretraining
         motivation and perceived transfer of training. They also determined the strong
         relationship between supervisor support and pretraining motivation. Peer sup-
         port was not significantly related to pretraining motivation, and subordinates
         and top management support were negatively related. Brinkerhoff and
         Montesino (1995) found a significant relationship between management
         support and transfer of training. Therefore, one can argue that the favorability
         of the transfer climate is contingent on several factors: peer support, supervi-
         sor support, availability of resources to apply new knowledge and skill (such
         as time, tools, and human resources), a rewards system in place with positive
         reinforcement, and negative reinforcement (sanctions, punishment).
              The set of influences described is defined as the transfer system. It is a
         much broader construct than transfer climate. Transfer climate alone does not
         constitute the whole set of influences on transfer of training. As Holton et al.
         (2000) state, it also includes training design, personal characteristics,
         opportunity to use training, and motivational influences. Therefore, training

                                               HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 392

         392                                                        Holton, Bates, Bookter, Yamkovenko

         transfer can be understood only by examining and evaluating the complete set
         of influences described above. At the same time, these factors have to be care-
         fully managed and fostered in organizations in order to implement training
         interventions successfully.
              Nevertheless, it is evident that often the analysis of transfer system is
         underused or even neglected, and not enough attention is given to measuring
         the constituents of the system. Georgenson (1982) estimates that of the expen-
         ditures for training and development in American industry, only 10 percent
         actually result in a transfer from the training to the job. Indirect costs of
         on-the-job training, combined with formal training costs, range between $200
         billion and $400 billion a year (Holton et al., 1997). Ultimately, because of
         poorly assessed training input factors, improperly designed training, and unfa-
         vorable training transfer environment, what is learned in training is rarely
         applied on the job.

         Methodology
         Two sets of procedures were used in this study for data collection. First, an
         extensive review of literature was conducted to identify a set of comparison
         measures against which to evaluate the convergent and divergent validity of the
         LTSI scales. Second, data were collected and analyzed in order to examine
         the correlations between the comparison measures and the LTSI scales.
              Learning Transfer Systems Inventory. The most recent version of the
         LTSI contains sixty-eight items, which are subdivided into two domains: Train-
         ing Specific and Training in General. The first domain contains forty-five items
         and the second domain twenty-three items. There are four sets of factors in the
         instrument: Motivation, Work Environment, Ability, and Secondary Influences.
         Overall factor analysis revealed sixteen factors. This version of LTSI also
         contains twenty-one research items, which are being tested for their validity.
         These items have been shown to work well in cross-cultural settings in studies
         described earlier.
              The LTSI originated after critical analysis of Kirkpatrick’s four-level evalua-
         tion model (Holton, 1996). Holton et al. (2000) propose a new model to evaluate
         training transfer specifically, which provides the theoretical frame for the
         instrument. Four domains are considered in the conceptual model of organi-
         zational performance improvement from training. Each domain represents a
         system of factors important to learning, individual performance, and, ulti-
         mately, organizational results:

             • Motivational factors are direct measures of transfer-related motivation
         (Motivation to Transfer) and two measures that stem from Vroom’s expectancy
         theory (Transfer Effort–Performance Expectations and Performance–Outcome
         Expectations) and are intended to assess transfer-related expectations.

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 393

         Validity of the Learning Transfer System Inventory                                       393

         Motivation to Transfer measures the extent to which individuals are motivated
         to use learning in their work and therefore plan to use new skills and knowl-
         edge which will help them perform more effectively on the job. Transfer
         Effort–Performance Expectations measures the extent to which individuals
         believe that applying new skills and knowledge learned will improve their
         performance. In turn, the Performance–Outcomes Expectations scale assesses
         the extent to which employees believe that applying new knowledge learned
         in training will lead to some kind of recognition valuable and meaningful
         to them.
              • Secondary influences, which can also be classified as trainee character-
         istics scales, are presented by Learner Readiness and Performance Self-Efficacy.
         Learner Readiness relates to the degree of the preparedness of the trainee to
         enter training, including knowing what to expect during training and under-
         standing how training is related to job and work performance. Performance
         Self-Efficacy is essentially the trainee’s belief that he or she will be able to use
         the learned material on the job to improve performance. In other words, this
         scale represents the confidence of the employee that he or she can overcome
         obstacles that hinder the use of new knowledge and skills at work.
              • Environmental elements or factors are measured by three scales that
         address employee-supervisor relationship: Supervisor Support for Transfer,
         Supervisor Sanctions, and Performance Feedback. Essentially these scales
         address managers’ involvement in clarifying performance expectations after
         training, identifying opportunities to use new knowledge and skills, setting
         realistic goals based on training, and working with individuals on problems
         encountered. On the negative side, Supervisor Sanctions indicate the degree
         of opposition to application of new skills and knowledge, lack of assistance to
         identify opportunities to use new skills, and providing negative or inadequate
         feedback when individuals successfully apply learning on the job. The Peer
         Support and Openness to Change scales assess the work-group-related factors
         that influence training transfer. The Peer Support scale aims to establish
         whether peers mutually implement opportunities to apply skills and knowl-
         edge learned in training, encourage each other to use new skills, and display
         patience and appreciation for the use of new skills. The Openness to Change
         scale addresses the extent to which work groups are willing to invest energy to
         change and provide support to individuals who use new techniques learned
         in training. The reward system in place in organizations and the rewards an
         employee expects for successful training completion and implementation of
         new knowledge and skills on the job are important constructs that influence
         the amount of transfer on the job. These factors are measured by two scales:
         Performance Outcomes Positive and Performance Outcomes Negative. Posi-
         tive outcomes delineated here include increased productivity at work,
         increased personal satisfaction, respect, increase in salary or other types of
         rewards, and promotion. Negative outcomes include reprimands, penalties,
         peer resentment, and lack of rewards.

                                                HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 394

         394                                                        Holton, Bates, Bookter, Yamkovenko

              • Ability elements have been discussed in transfer research as some of the
         most important elements affecting transfer of training on the job. The ability
         to apply learning to the job setting is addressed by the Opportunity to Use
         Learning scale and the Personal Capacity for Transfer scale. The Opportunity
         to Use Learning scale is designed to measure the extent to which an employee
         is given the opportunity to apply what he or she learned during training in
         terms of adequate equipment, information, human and financial resources,
         materials, and supplies. The Personal Capacity for Transfer assesses the extent
         to which individuals’ workload, schedule, personal energy, and stress level
         facilitate or inhibit the application of new knowledge and skills. It is impor-
         tant that training be consistent with job requirements and skills and knowl-
         edge taught in training be similar to performance expectations. The adequacy
         of these elements is measured by the Perceived Content Validity scale. Finally,
         the Transfer Design scale measures the extent to which training has been
         designed to clearly link learning with on-the-job performance through the use
         of clear examples, methods, and activities.

              Procedures: Research Question 1. To address research question 1, an
         extensive systematic search of literature was conducted to identify comparison
         measures for the LTSI factors that were used in this study. The review included
         resources such as relevant journals from a variety of disciplines, books, and
         other literature using the Educational Resources Information Center, Disserta-
         tion Abstracts International, the Library Information System, and ABI/Inform
         (produced by University Microfilms International). Search terms thought to
         relate to the learning transfer climate constructs were identified: learning cli-
         mate, transfer environment, business climate, organization environment, psycholog-
         ical climate, work environment, organizational climate, organizational culture,
         learning culture, and organizational behavior.
              The goal of our review was to find a set of psychometrically sound
         comparison scales from those available either in the research literature or com-
         mercially. Because there were no other measures of learning transfer climate
         available when this study was conducted, comparison measures had to be
         drawn from a variety of instruments that measured related but different
         constructs. The procedures for identifying comparison measures were based
         on instrument development criteria outlined by Robinson, Shaver, and
         Wrightsman (1991) and Nunnally and Bernstein (1994). These criteria
         were used as a screening device to narrow the large volume of potential
         measures. The goal was to find instruments or measures that best met these
         criteria. Table 1 shows the criteria used for identifying and screening possible
         comparison instruments and the steps used for each criterion.
              Next, the psychometric qualities of the instruments that passed the first
         screen were assessed. The adequacy of measures was evaluated with the
         Robinson et al. (1991) framework. The framework involved examining
         the theoretical development structure, available scale norms, inter-item

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 395

         Validity of the Learning Transfer System Inventory                                             395

          Table 1. Criteria and Steps Used in Identifying Candidate Measures for
                              Comparison with LTSI Factors
         Criteria                                                               Steps
         1. The purpose and intended use of             1. A thorough review of the factors
         candidate measures were examined to            comprising the LTSI instrument was made
         determine similarities or differences from     (for example, definitions, examples, and
         the measure of interest.                       researcher-perceived inferences).
                                                        2. Instruments developed for the purpose of
                                                        measuring a variety of factors related to the
                                                        learning transfer climate (for example,
                                                        learning climates, business climate, transfer
                                                        environment, organizational behavior, work
                                                        environment, and psychological climate)
                                                        were considered.
                                                        3. Instruments that were intended for use in
                                                        a variety of organizational work
                                                        environments with a wide variety of
                                                        employees were considered.
                                                        4. Instruments that were intended to
                                                        measure employee perceptions of the
                                                        training and work environment as opposed
                                                        to how they desire the workplace to be were
                                                        considered.
                                                        5. Instruments that were intended to analyze
                                                        individual-level responses, as opposed to
                                                        group- or unit-level analysis of responses,
                                                        were considered.
                                                        6. Instruments that were self-administered
                                                        and did not need outside facilitation were
                                                        considered.

         2. Instruments with scales and scale items     1. The constructs, classification schemes,
         related to the items in the measure of         dimensions, scales, and scale items for
         interest were candidates.                      potential comparison measures were
                                                        examined to logically determine similarities
                                                        to or differences from the factors in the
                                                        nomological network of the LTSI.
                                                        2. Scales that had twenty or fewer items were
                                                        considered for the study. The length of the
                                                        instrumentation used in the study was a
                                                        major factor.
                                                        3. Scale items with clear, simple language
                                                        were considered because the instrumentation
                                                        is to be administered to a wide variety of
                                                        employees of varying educational levels.
                                                                                                (Continued)

                                                      HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 396

         396                                                        Holton, Bates, Bookter, Yamkovenko

                                              Table 1. (Continued)
         Criteria                                                                Steps
         3. Developmental studies for the                    1. Measures with data generated with
         candidate measures were examined to                 samples similar to those used for the
         determine the methodology used.                     development of the LTSI were used.
                                                             Instrument development studies using
                                                             individuals working in companies or
                                                             organizations in the public and private
                                                             sectors were considered.
                                                             2. A review of the techniques (survey,
                                                             interviews, and observation) used in the
                                                             developmental studies was made to determine
                                                             similarities to or differences from the LTSI.
                                                             The LTSI used the survey technique for data
                                                             collection. Therefore, only instruments with
                                                             questionnaires were used in this study.

         correlations, coefficient alphas, factor analysis results, test-retest reliability, con-
         vergent validity, and divergent validity. Using a four-point scale (4, Exemplary;
         3, Extensive; 2, Moderate, 1; Minimal, 0, None), the researchers rated each
         potential comparison scale against these criteria. Few instruments were found
         that fully met all of the criteria. Thus, the final selection instruments often
         required balancing a trade-off of theoretical, practical, and psychometric issues.
         This was due in part to the lack of the attention to measurement issues in past
         research.
              For practical reasons of instrument length, only two scales were selected
         for comparison with each LTSI scale used in the study. This approach reduced
         the number of items used in the instruments to a manageable level. In all, forty-
         two scales were used in this study, including fourteen of the sixteen LTSI scales
         and twenty-eight comparison measures. Two of the LTSI scales were excluded
         from this analysis because suitable comparison measures could not be found.
              Procedures: Research Question 2. This section describes the research pro-
         cedures used to assess the relationship between the comparison measures and
         the LTSI scales.
              Sample. Two hundred thirty-seven participants in training programs from
         a large quasi-public organization completed the instruments. The respondents,
         in five training satellite offices from five states, were from the Midwest district.
         The employees represented a variety of job levels in the organization: craft
         employees such as clerks and carriers, first-line supervisors, and middle
         management (for example, managers and postmasters). Data were collected
         during training sessions or programs in which subjects participated. The
         training sessions and programs ranged from 4 hours to 104 hours, with only
         four participants completing the 104-hour training session and fourteen
         completing 80 hours of training. The majority of the training ranged from 4 to
         32 hours.

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 397

         Validity of the Learning Transfer System Inventory                                      397

               Participants returned 204 usable sets of data. This provided a sample with
         adequate power to detect significant correlations of .20 or higher. Correlations
         below .20 were considered too small to be meaningful in defining the nomo-
         logical network of constructs related to the LTSI.
               Measurement Instruments. The fourteen LTSI and twenty-eight compari-
         son scales yielded 322 items. Because of the large number of items and
         concerns about respondent fatigue, the items were divided into two survey
         instruments: one was administered at the beginning of the training class
         and the other at the end of the class. All of the LTSI scales were in the
         second administration because it is designed to be administered at the end
         of training. The first survey instrument, administered at the beginning of
         training, contained only comparison scales. To balance the number of items
         between the two instruments, some of the comparison scales were included
         with the LTSI items in the second survey administered at the end of the
         training class. Table 2 shows the LTSI scales, their definition, a sample item
         from each scale, and the reliabilities of each scale.
               Because of the time span between completing the first and second instru-
         ments, there was a concern that the training might create an attitude change
         that would influence trainee responses to the two instruments. Therefore, an
         effort was made to statistically control for mood change between the beginning
         and the end of the training sessions. The Positive and Negative Affect Sched-
         ule (PANAS; Watson, Clark, & Tellegan, 1988) was added to assess any mood
         shifts that participants experienced between the beginning and the end of the
         training session that might have influenced the data. The PANAS consisted
         of twenty terms (ten positive and ten negative) rated along a five-point Likert-
         type scale (1  very slightly or not at all, 2  a little, 3  moderately, 4 
         quite a bit, and 5  extremely). Respondents were asked the extent to which
         they experienced each mood state during a specific time frame. For this study,
         respondents were asked to indicate the “extent you feel this way right now, that
         is, at the present moment.” The “present moment” time frame was used to cap-
         ture state affectivity at the beginning and end of training. Dependent t-tests
         were calculated to determine if the means for the beginning and ending scores
         differed significantly, suggesting a shift in the mood of the participants between
         the beginning and end of the sessions. Pearson’s correlations were also calcu-
         lated for PANAS scales to examine the individual differences in the positive and
         negative scales.
               Administrative Procedures. Training specialists for the Midwest district
         were administered the questionnaires. They were provided a prepared script
         explaining the purpose of the study and directions for completing the
         questionnaires. The script was read to the participants to ensure consistency
         of the message. Completed questionnaires were returned to the researchers
         for processing.
               Data Analysis. Pearson’s product moment correlation statistic was used
         in most cases to examine the associations between the LTSI scales and the

                                               HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
Table 2. LTSI Scale Definitions and Sample Items
                                                                                                                         Number
Factor                                        Definition                                     Sample Item                 of Items    a

Training-specific scales
 1. Learner Readiness        Extent to which individuals are prepared       Before the training I had a good                4       .73
                             to enter and participate in training.          understanding of how it would fit my
                                                                            job-related development.
 2. Motivation to Transfer   Direction, intensity, and persistence of       I get excited when I think about trying to      4       .83
                             effort toward using skills and knowledge       use my new learning on my job.
                             learned in a work setting.
 3. Personal Outcomes—       Degree to which applying training on the       Employees in this organization receive          3       .69
                                                                                                                                          hrdq18306.qxd 8/4/07 2:46 PM Page 398

    Positive                 job leads to positive outcomes for the         various perks when they utilize newly
                             individual.                                    learned skills on the job.
 4. Personal Outcomes—       Extent to which individuals believe that       If I do not utilize my training I will be       4       .76
    Negative                 not applying skills and knowledge learned      cautioned about it.
                             in training will lead to negative personal
                             outcomes.
 5. Personal Capacity for    Extent to which individuals have the time,     My workload allows me time to try the           4       .68
    Transfer                 energy, and mental space in their work         new things I have learned.
                             lives to make changes required to transfer
                             learning to the job.
 6. Peer Support             Extent to which peers reinforce and            My colleagues encourage me to use the           4       .83
                             support use of learning on the job.            skills I have learned in training.
 7. Supervisor Support       Extent to which supervisors or managers        My supervisor sets goals for me which           6       .91
                             support and reinforce the use of training on   encourage me to apply my training on
                             the job.                                       the job.
 8. Supervisor Sanctions     Extent to which individuals perceive           My supervisor opposes the use of the            3       .63
                             negative responses from supervisors or         techniques I learned in training.
                             managers when applying skills learned in
                             training.
9. Perceived Content Validity   Extent to which trainees judge training         What is taught in training closely matches   5   .84
                                 content to accurately reflect job               my job requirements.
                                 requirements.
10. Transfer Design              Degree to which (1) training has been           The activities and exercises the trainers    4   .85
                                 designed and delivered to give trainees the     used helped me know how to apply my
                                 ability to transfer learning to the job and     learning on the job.
                                 (2) training instructions match job
                                 requirements.
11. Opportunity to Use           Extent to which trainees are provided with      The resources I need to use what I learned   4   .70
                                 or obtain resources and tasks on the job        will be available to me after training.
                                 enabling them to use training on the job.
                                                                                                                                        hrdq18306.qxd 8/4/07 2:46 PM Page 399

General scales
12. Transfer Effort–             Expectation that effort devoted to              My job performance improves when I use       4   .81
    Performance Expectations     transferring learning will lead to changes in   new things that I have learned.
                                 job performance.
13. Performance–Outcomes         Expectation that changes in job                 When I do things to improve my               5   .83
    Expectations                 performance will lead to valued outcomes.       performance, good things happen to me.
14. Resistance to Change         Extent to which prevailing group norms          People in my group are open to changing      6   .85
                                 are perceived by individuals to resist or       the way they do things.
                                 discourage the use of skills and knowledge
                                 acquired in training.
15. Performance Self-Efficacy    An individual’s general belief that he or she   I am confident in my ability to use newly    4   .76
                                 is able to change performance when he or        learned skills on the job.
                                 she wants to.
16. Performance Coaching         Formal and informal indicators from an          After training, I get feedback from people   4   .70
                                 organization about an individual’s job          about how well I am applying what I
                                 performance.                                    learned.
hrdq18306.qxd 8/4/07 2:46 PM Page 400

         400                                                        Holton, Bates, Bookter, Yamkovenko

         comparison measures. However, three of the comparison measures consisted
         of dichotomous data requiring a true-false response. For these comparisons,
         the point biserial correlation was used. The following framework was used
         to describe the magnitude of correlation:

         •   .00 to .19, negligible association
         •   .20 to .49, low association
         •   .50 to .69, moderate association
         •   .70 to .85, high association
         •   .86 to 1.00, very high association (Ary, Jacobs, & Razavieh, 1996)

              Pearson’s partial correlations measure the strength and direction of a
         relationship between two variables while controlling for the effect of one or more
         additional values (Nunnally & Bernstein, 1994). For this study, partial correla-
         tions are used for any comparison between a scale on the instrument adminis-
         tered at the beginning of training, partialing out the effect of mood change as
         measured by the PANAS. For comparison scales included on the same survey
         instrument as the LTSI factors, a simple Pearson correlation was sufficient.

         Results
         This section reports the results of the two research questions.
              Research Question 1: Comparison Measures Selected for Use. Of the
         fifty-two scales investigated for use in this study, seventeen were selected for
         use. This section describes the scales selected and the rationale underlying the
         selection of each scale.
              All of the instruments except the self-efficacy scale (Sherer et al., 1982)
         met the first set of criteria (see Table 1). Data used in the initial development
         of this instrument were drawn from students in a college-level psychology
         class, not individuals working in organizations. For our purposes, this was not
         ideal, to the extent we were looking for instruments that evidenced some
         generalizability to organizational setting, However, given the nature of the
         construct, self-efficacy, which examines an individual’s past experiences with
         success and failure in a variety of situations, including work situations; the
         apparent relatedness of the scales items to LTSI 15: Performance Self-Efficacy;
         and the fact that two sets of factor analysis data were collected to verify factor
         structure, the instrument was determined appropriate for the study. In addi-
         tion, another study during the developmental process (Sherer et al., 1982)
         included a sample similar to that of the LTSI.
              Second, the instruments in this study were rated based on the components
         of the instrument evaluation model (Robinson et al., 1991). Because many
         studies have not used psychometrically valid measurements, only two of the
         comparison measures, the Job Descriptive Index and KEYS Environment Scale,
         met to some degree each of the eight criteria of the model. In this rating

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 401

         Validity of the Learning Transfer System Inventory                                      401

         methodology, three of the instruments were rated in the moderate category:
         Leader Reward Behavior Scale, Work Related Expectancies Scale, and Manifest
         Need Questionnaire. The remaining fourteen instruments received ratings in
         the “extensive” category for evaluating instruments. Each of the instruments
         and selected scales is described next.
              Work Environment Scale. The Work Environment Scale (WES; Moos,
         1994) is a ninety-item instrument developed to measure the social environ-
         ment of work settings along three broad dimensions (relationships, personal
         growth, and systems maintenance and change) assessed by ten scales. The
         format for the WES requires true-false responses to each statement.
              Moos (1994) used a sample of 1,045 employees (retail food industry, office
         managers, clerical workers, radio station workers, employees in education, and
         health care workers) to examine internal consistency reliabilities for the ten
         scales. The results indicated that reliability scores were in an acceptable range
         (involvement, a  .84; coworker cohesion, a  .69; supervisor support, a 
         .77; autonomy, a  .73; task orientation, a  .76; work pressure, a  .80;
         clarity, a  .79; managerial control, a  .76; innovation, a  .86; and
         physical comfort, a  .81).
              Three WES scales were used in this study. The Coworker Cohesion scale of
         WES was used as a comparison measure with LTSI 6: Peer Support, because
         both scales examined interaction among coworkers. The Work Pressure scale of
         WES was compared to the LTSI 5: Personal Capacity for Transfer, because both
         measures addressed time as it related to the demands of the job. Finally, the
         Managerial Control scale was correlated with the LTSI 4: Personal Outcomes—
         Negative, because both assessed perceived managerial impact on employee
         performance.
              Job Descriptive Index. Based on job satisfaction theory, the Job
         Descriptive Index ( JDI; Smith, Kendall, & Hulin, 1969) was developed to
         assess an individual’s job satisfaction. It consists of ninety items with five
         scales intended to reflect the idea that job satisfaction factors (work, pay,
         promotion, supervision, and coworkers) are principal components in worker
         motivation and behavior (Balzer et al., 1997). The JDI was revised in 1985
         to accommodate workplace changes but still assessed the five original
         constructs.
              A random sampling procedure, stratified by state population, was used to
         establish normative data for the JDI (Smith-Jentsch, Jentsch, Payne, & Salas,
         1996). A sample of working individuals was obtained from the 1990 U.S. Cen-
         sus and social security database. Internal reliability estimates for each of the
         five scales were calculated with the sixteen hundred cases from the national
         norm database. Cronbach’s alpha coefficient estimates of reliability for each of
         the scales were as follows: Work  .90; Pay  .86; Opportunities for Promo-
         tion  .87; Supervision  .91; and Co-workers  .91.
              Two comparison scales were selected from the JDI. The Opportunity for
         Promotion was compared with the LTSI 3: Personal Outcomes—Positive,

                                               HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 402

         402                                                        Holton, Bates, Bookter, Yamkovenko

         because both scales address employee expectations of rewards from their
         employer. The Supervision scale was compared with LTSI 7: Supervisor Support,
         because both scales address a supervisor’s impact on employee job performance.
              KEYS Environmental Scales. KEYS Environmental Scales (KEYS; Amabile,
         Burnside, & Gryskiewicz, 1995) was developed to assess perceived stimulants
         and obstacles to creativity in organizational work environments. It is based on
         creativity and innovation theory and sees three general organization components
         (organizational motivation to innovate, resources, and management practices) as
         key in the innovation process. KEYS comprises seventy-eight items with ten
         scales that focus on stimulants to creativity, obstacles to creativity, and outcomes.
         It uses a four-point, Likert-type response scale.
              Validation studies of KEYS were conducted using 12,525 respondents
         (Amabile, 1988; Amabile, Gryskiewicz, Burnside, & Koester, 1990). The respon-
         dents represented employees in a variety of functions and departments from over
         twenty-six organizations (industrial, high technology, biotechnology, electronics,
         health, and pharmaceutical products). The findings showed that KEYS had “an
         acceptable factor structure; the median scale reliability was .84; all items corre-
         lated more strongly with their own scale than they did with any other scale; and
         test-retest reliability over a three month period was above .70” (Amabile et al.,
         1990, p. 26). Correlation studies comparing KEYS with the Work Preference
         Inventory, a personality measure of motivational orientation, and the Kirton-
         Adaption-Innovation Inventory, a cognitive style measure, showed relatively
         low correlations, suggesting divergent validity and “that respondents’ rating of
         their work environment are not merely reflections of their own personal char-
         acteristics” (Amabile et al., 1990, p. 26). KEYS correlated only moderately
         with the WES, indicating that the measures assess different aspects of the work
         environment.
              Two KEYS scales were selected as comparison measures in our study: Supervi-
         sory Encouragement of Creativity and Work Group Support. The Supervisory
         Encouragement of Creativity scale was used as a comparison to LTSI 7: Super-
         visor Support, because both appear to address the effect of supervisors’ man-
         agement styles on employee performance. The Work Group Support scale was
         correlated with LTSI 14: Resistance to Change, because both scales address
         aspects of group support for change.
              Perceived Work Environment. Perceived Work Environment (PWE; Newman,
         1977) was developed as a comprehensive measure of employee perceptions
         of the work environment. It was designed to assess the state of a given work
         environment and evaluate the effect of programs aimed at modifying
         organized work environments (Newman, 1977). The PWE consists of
         sixty items with eleven scales and is formatted on a five-point Likert scale
         (never true, almost never true, sometimes true, almost always true, and
         always true).
              Validation data for the PWE are based on five studies conducted with
         samples representing a diversity of organizational levels and functional

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 403

         Validity of the Learning Transfer System Inventory                                     403

         departments (executive, accounting, personnel, records, administration ser-
         vices, underwriting, claims, and policyholder services). The initial study used
         a sample of 710 employees from a Midwest regional office of a large multiline
         insurance company. The results indicated acceptable levels of reliability
         (median range .71) for all of the scales with the exception of the job
         responsibility/importance scale. The reliability scores were similar in the four
         replication studies, indicating the same underlying dimensional structure
         (Newman, 1977). A comparison of the factor solutions from the five data col-
         lections reported that the factor congruence coefficients for the eleven-factor
         structure ranged from .45 to .95, with a median of .82. The PWE was corre-
         lated with the Job in General scale (Smith et al., 1969) with coefficients rang-
         ing from .10 to .40, indicating that the PWE scales “are not just another
         measure of job satisfaction” (Newman, 1977, p. 529).
              Five PWE scales were used in this study: Employee Competence,
         Employee Work Motivation, Coworker Relations, Task Characteristics and
         Pressure to Produce. Employee Competence was correlated with the LTSI 1:
         Learner Readiness scale, because both examined employee readiness.
         Employee Work Motivation was compared with the LTSI 2: Motivation to
         Transfer, because both scales address dimensions of employee motivation.
         Coworker Relations was correlated with the LTSI 6: Peer Support, because
         both examined how employees interact with each other. Task Characteristics
         was compared to LTSI 11: Opportunity to Use Learning, because both scales
         addressed characteristics of employee job tasks. Pressure to Produce was com-
         pared to LTSI 5: Personal Capacity for Transfer, because both scales examined
         work pressure on the job.
              Internal Work Motivation Scale. The Internal Work Motivation Scale
         (IWMS; Hackman & Oldham, 1975), based on work redesign theory, was
         constructed to assess the degree to which employees are self-motivated to
         perform effectively. IWMS is a six-item scale with a seven-point Likert
         response format.
               A validation study (Hackman & Oldham, 1975) using 658 employees in
         seven organizations reported a Spearman-Brown internal reliability coefficient
         of .76. Within this sample, the IWMS measure was correlated with the General
         Job Satisfaction scale (r  .51), the Growth Satisfaction scale (r  .56), and
         the Experience Meaningfulness of Work scale (r .66).
              The Internal Work Motivation Scale was compared to LTSI 2: Motivation
         to Transfer scale, on the basis that both measures examined motivational
         factors affecting employee job performance.
              Index of Organizational Reactions. The Index of Organizational Reactions
         (IOR; Smith, 1976), based in job satisfaction theory, was developed to
         measure multiple facets of employee job satisfaction and perceived
         relationships between job features and work performance (Smith, 1976). The
         IOR comprises forty-two items with eight scales and has been used with
         either five- or six-point, Likert-type response scales.

                                              HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
hrdq18306.qxd 8/4/07 2:46 PM Page 404

         404                                                        Holton, Bates, Bookter, Yamkovenko

              Validation studies (Smith, 1976) were conducted using five samples made
         up of 12,971 employees from numerous locations, work functions, and depart-
         ments within Sears, Roebuck and Company. In this study, the IOR scales
         showed acceptable internal reliability estimates: supervision, a  .90;
         company identification, a .82; kind of work, a  .89; amount of work,
         a  .77; coworkers, a  .77; physical conditions, a  .90; financial rewards,
         a  .85; and career future, a  .83. Principal component factor analysis
         produced virtually identical factor structures across five samples of workers in
         studies conducted over a three-year period.
              The Financial Elements and Supervision scales of the IOR were used in
         this study. The Financial Elements scale was used as a comparison scale for
         LTSI 3: Personal Outcomes–Positive, because both scales addressed employee
         expectations for rewards. The IOR Supervision scale was compared to LTSI 8:
         Supervisor Sanctions, because both scales tap perceptions about how negative
         reactions from supervisors influence employee job performance.
              Leader Reward Behavior Scale. The Leader Reward Behavior Scale (LRBS;
         Sims & Szilagyi, 1975) was drawn from leadership theory and was designed
         to measure the extent to which a subordinate perceives that positive or
         negative rewards received through his or her supervisor reflect his or her job
         performance. It contains two scales: Positive Reward Behavior and Punitive
         Reward Behavior. The LRBS uses a seven-point, Likert-type response scale.
              Internal reliability estimates, based on data from 630 paramedical and
         support personnel in a university medical center, for the two scales
         were acceptable: .93 for Positive Reward Behavior and .70 for Punitive
         Reward Behavior (Sims & Szilagyi, 1975). Factor analysis of data from 192
         managerial, engineering, and supervisory manufacturing employees
         confirmed the two-factor structure of the LRBS, with congruency coefficients
         of .95 for Positive Reward Behavior and .91 for Punitive Reward Behavior.
         Spearman-Brown internal reliabilities were .92 and .88, respectively (Keller &
         Szilagyi, 1978).
              Our study used the Punitive Reward Behavior scale as a comparison for
         LTSI 8: Supervisor Sanctions. Both scales examine the effect of perceived
         negative supervisor actions (or reactions) on employee job performance.
              Facet-Specific Job Satisfaction. The Facet-Specific Job Satisfaction (FSJS;
         Quinn & Staines, 1979) was designed to measure a worker’s evaluation of
         his or her job. The instrument contains thirty-three items measuring six
         scales. The conceptual framework is based in job satisfaction theory.
              A validation study by Quinn and Staines (1979) used a sample of 1,515
         respondents designed to be representative of all employed adults, industries,
         and occupations in the United States. Initial evidence of the validity of the
         six-factor structure was derived through factor analysis. Subsequent reliability
         estimates for the scales included comfort (a  .69), challenge (a  .88), finan-
         cial rewards (a  .66), relations with coworkers (a  .61), resource adequacy
         (a  .88), and promotions (a  .76).

         HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq
You can also read